Sparse DNNs with Improved Adversarial Robustness

10/23/2018
by   Yiwen Guo, et al.
0

Deep neural networks (DNNs) are computationally/memory-intensive and vulnerable to adversarial attacks, making them prohibitive in some real-world applications. By converting dense models into sparse ones, pruning appears to be a promising solution to reducing the computation/memory cost. This paper studies classification models, especially DNN-based ones, to demonstrate that there exists intrinsic relationships between their sparsity and adversarial robustness. Our analyses reveal, both theoretically and empirically, that nonlinear DNN-based classifiers behave differently under l_2 attacks from some linear ones. We further demonstrate that an appropriately higher model sparsity implies better robustness of nonlinear DNNs, whereas over-sparsified models can be more difficult to resist adversarial examples.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
03/13/2021

Learning Defense Transformers for Counterattacking Adversarial Examples

Deep neural networks (DNNs) are vulnerable to adversarial examples with ...
research
04/21/2020

EMPIR: Ensembles of Mixed Precision Deep Networks for Increased Robustness against Adversarial Attacks

Ensuring robustness of Deep Neural Networks (DNNs) is crucial to their a...
research
06/11/2021

Knowledge Enhanced Machine Learning Pipeline against Diverse Adversarial Attacks

Despite the great successes achieved by deep neural networks (DNNs), rec...
research
03/22/2021

Fast Approximate Spectral Normalization for Robust Deep Neural Networks

Deep neural networks (DNNs) play an important role in machine learning d...
research
05/20/2022

Robust Sensible Adversarial Learning of Deep Neural Networks for Image Classification

The idea of robustness is central and critical to modern statistical ana...
research
12/24/2021

CatchBackdoor: Backdoor Testing by Critical Trojan Neural Path Identification via Differential Fuzzing

The success of deep neural networks (DNNs) in real-world applications ha...
research
05/31/2019

L0 Regularization Based Neural Network Design and Compression

We consider complexity of Deep Neural Networks (DNNs) and their associat...

Please sign up or login with your details

Forgot password? Click here to reset